\begin{table}[h!]
    \centering
    \caption{Incidence Analysis of 421-a: Actual Versus Incidence-Minimizing Counterfactual}
    \label{tab:incidence}
    \small
    \begin{tabular*}{\textwidth}{@{\extracolsep{\fill}}lcc@{}}
\toprule
 & (1) & (2)                  \\ 
                & Actual       & Counterfactual       \\ \midrule \addlinespace
Developer Incidence Share           & $$incidence_estimate$$***    & $$incidence_cf_estimate$$***      \\
                         & ($$incidence_stderr$$)    & ($$incidence_cf_stderr$$)    \\ \addlinespace 
$N$                       & $$incidence_obs$$    & $$incidence_cf_obs$$   \\
Clusters & $$incidence_clust$$    & $$incidence_cf_clust$$    \\\addlinespace \bottomrule
\end{tabular*}
      

        	\begin{spacing}{1}
\begin{tablenotes}
      \item \footnotesize \textit{Notes:} This table presents estimates of the share of the total citywide fiscal cost of 421-a captured by developers as profits. The left column contains the estimated developer incidence share under the actual 421-a program, whereas the right column contains the developer incidence share under a counterfactual program that minimizes developer incidence subject to a constraint that it achieves the same take-up rate. Standard errors are computed by a cluster-bootstrap at the level of Neighborhood Tabulation Areas. $\sym{*} = p < 0.10$, $\sym{**} = p < 0.05$, $\sym{***} = p < 0.01$.
      \end{tablenotes}
       \end{spacing}
\end{table}